The present invention pertains to the field of digital computer systems. More particularly, the present invention relates to thermal management in the operational of thin servers.
Digital computers are being used today to perform a wide variety of tasks. Many different areas of business, industry, government, education, entertainment, and most recently, the home, are tapping into the enormous and rapidly growing list of applications developed for today""s increasingly powerful computer devices. The widespread deployment of computer systems has led to the equally widespread deployment of interconnected networks of digital computer systems (e.g., LANs, WANs, Intranets, the Internet, etc.). The infrastructure of network connected digital computer systems has led to the ubiquitous adoption of network based applications, Web applications, Web based services, and the like. Many types of these network-based applications rely upon network-connected servers for their functionality.
With the increasing use of network based applications, there is interest in increasing the performance of these applications while simultaneously decreasing cost of the hardware infrastructure required to support these applications. Specifically, thin servers are being widely adopted as a means of providing Web/network based services cost effectively.
Generally, the term xe2x80x9cthin serverxe2x80x9d refers to a network-based computer specialized for some particular function such as, for example, a print server, DSL router, or network attached storage (NAS). Thin servers are particularly designed for ease of installation. Thin servers generally have very little expandability and are deployed without a keyboard or monitor. Web server software is typically built in allowing management and control via a Web browser residing on a client platform in the network.
The thin server term has also come to refer to the size of such network based computers. Thin servers are typically designed to be rack mounted. As thin servers have become more specialized and more developed, development effort has been expended in making the servers consume as little space as possible within a server rack. Contemporary thin servers can be stacked on top of each other and take up considerably less space than, for example, tower cases.
To improve performance, the industry trend for computer servers is to put higher power (e.g., higher performance) chips in smaller server chassis. Thus, a larger number of servers (e.g., thin servers) can be deployed within the available space of a given server rack. As the server chassis become increasingly thin (e.g. smaller), a direct product of this trend is that less height within the chassis is available for heat sinks of high-powered chips. This problem is expressly evident in thin servers such as 1 U high servers where only 1.75 inches of height is available for the entire system chassis.
Prior art FIG. 1 shows a vertical cross section of a traditional prior art heat sink where the ability of the heat sink to cool is a function of the surface area of the heat sink fins (among other factors). As depicted in FIG. 1, the heat sink comprises the heat sink base 118 and the heat sink fins 101. The heat sink fins 101 are coupled to the base of the heat sink 118. A thermal interface 116 connects the base 118 to a chip lid 114 (e.g., the upper surface of the chip). An interconnect 112 electrically couples the chip to a printed circuit board 110 within the server. As can be seen in FIG. 1, the height of the fins 101 of the heat sink is limited in a thin server. The fins 101 must dissipate the majority of the heat dissipated by the chip. As can be seen in FIG. 1, the smaller the chassis of the server, the less available surface area for heat dissipation is provided by the fins 101. This can have extremely harmful effects on the longevity and the performance of high-powered chips.
Prior art solutions to this problem have not been satisfactory. One prior art solution involves making a wider heat sink in order to allow more fins and thus more surface area for cooling. A disadvantage with this solution is that the thermal resistance increases for the fins that are farther away from the heat source. Thus diminishing returns are realized for the added fins towards the outer edges of the width-expanded heat sink. Another disadvantage of the solution is that the wider heat sink consumes more PCB area that could otherwise be used for other components.
Another prior art solution involves using more expensive materials (i.e. higher thermal conductivity copper instead of aluminum) for the heat sink, or using heat pipes or vapor chambers in a conventional heat sink configuration, to increase cooling efficiency and to provide a somewhat higher degree of cooling with less fin area. A disadvantage with this solution is that it increases costs, and there are limits to the effectiveness of these methods.
Yet another prior art solution involves simply using lower power, lower performance processors and chips. Obviously, the disadvantage in this solution is the fact that the performance of the thin server, or any type of server for that matter, is adversely impacted.
Thus what is required is a solution that efficiently implements cooling for servers. What is required is a solution that efficiently manages thermal energy generated by the use of high performance, high-powered processors and chips. The required solution should efficiently dissipate thermal energy within the constraints of a thin server chassis. The present invention provides a novel solution to the above requirements.
A thermal dissipation apparatus for implementing chassis conducted cooling for a server is disclosed. In one embodiment, the apparatus includes a heat sink having a first surface and a second surface. The first surface is adapted to accept a chip thermal interface for a chip. The second surface is adapted to accept a chassis thermal interface for a chassis surface. The second surface implements a thermal conductive path from the chip to the chassis surface.